Implement hsaKmtGetTileConfig in thunk

Change-Id: Iba8d8efa46e3c268a03442d3db568e1b19230e94


[ROCm/ROCR-Runtime commit: 8351b3d2e8]

projects/rocr-runtime/include/hsakmt.h

@@ -651,6 +651,16 @@ hsaKmtSetTrapHandler(
     HSAuint64           TrapBufferSizeInBytes     //IN
     );
 
+/**
+  Gets image tile configuration.
+ */
+HSAKMT_STATUS
+HSAKMTAPI
+hsaKmtGetTileConfig(
+    HSAuint32           NodeId,     // IN
+    HsaGpuTileConfig*   config      // IN & OUT
+    );
+
 #ifdef __cplusplus
 }   //extern "C"
 #endif

projects/rocr-runtime/include/hsakmttypes.h

@@ -967,6 +967,21 @@ typedef struct _HsaPmcTraceRoot
     HSATraceId                  TraceId;
 } HsaPmcTraceRoot;
 
+typedef struct _HsaGpuTileConfig
+{
+    const HSAuint32 *TileConfig;
+    const HSAuint32 *MacroTileConfig;
+    HSAuint32 NumTileConfigs;
+    HSAuint32 NumMacroTileConfigs;
+
+    HSAuint32 GbAddrConfig;
+
+    HSAuint32 NumBanks;
+    HSAuint32 NumRanks;
+    /* 9 dwords on 64-bit system */
+    HSAuint32 Reserved[7]; /* Round up to 16 dwords for future extension */
+} HsaGpuTileConfig;
+
 #pragma pack(pop, hsakmttypes_h)
 
 

projects/rocr-runtime/include/linux/kfd_ioctl.h

@@ -364,6 +364,29 @@ struct kfd_ioctl_import_dmabuf_args {
 	uint32_t dmabuf_fd;	/* to KFD */
 };
 
+struct kfd_ioctl_get_tile_config_args {
+	/* to KFD: pointer to tile array */
+	uint64_t tile_config_ptr;
+	/* to KFD: pointer to macro tile array */
+	uint64_t macro_tile_config_ptr;
+	/* to KFD: array size allocated by user mode
+	 * from KFD: array size filled by kernel
+	 */
+	uint32_t num_tile_configs;
+	/* to KFD: array size allocated by user mode
+	 * from KFD: array size filled by kernel
+	 */
+	uint32_t num_macro_tile_configs;
+
+	uint32_t gpu_id;		/* to KFD */
+	uint32_t gb_addr_config;	/* from KFD */
+	uint32_t num_banks;		/* from KFD */
+	uint32_t num_ranks;		/* from KFD */
+	/* struct size can be extended later if needed
+	 * without breaking ABI compatibility
+	 */
+};
+
 #define AMDKFD_IOCTL_BASE 'K'
 #define AMDKFD_IO(nr)			_IO(AMDKFD_IOCTL_BASE, nr)
 #define AMDKFD_IOR(nr, type)		_IOR(AMDKFD_IOCTL_BASE, nr, type)
@@ -466,7 +489,10 @@ struct kfd_ioctl_import_dmabuf_args {
 #define AMDKFD_IOC_IMPORT_DMABUF		\
 	AMDKFD_IOWR(0x20, struct kfd_ioctl_import_dmabuf_args)
 
+#define AMDKFD_IOC_GET_TILE_CONFIG		\
+	AMDKFD_IOWR(0x21, struct kfd_ioctl_get_tile_config_args)
+
 #define AMDKFD_COMMAND_START		0x01
-#define AMDKFD_COMMAND_END		0x21
+#define AMDKFD_COMMAND_END		0x22
 
 #endif

projects/rocr-runtime/src/libhsakmt.ver

@@ -47,6 +47,7 @@ hsaKmtPmcStopTrace;
 hsaKmtMapGraphicHandle;
 hsaKmtUnmapGraphicHandle;
 hsaKmtSetTrapHandler;
+hsaKmtGetTileConfig;
 
 local: *;
 };

projects/rocr-runtime/src/memory.c

@@ -386,3 +386,39 @@ hsaKmtUnmapGraphicHandle(
 
 	return hsaKmtUnmapMemoryToGPU(PORT_UINT64_TO_VPTR(FlatMemoryAddress));
 }
+
+HSAKMT_STATUS
+HSAKMTAPI
+hsaKmtGetTileConfig(
+	HSAuint32		NodeId,	/* IN */
+	HsaGpuTileConfig	*config	/* IN & OUT */
+)
+{
+	struct kfd_ioctl_get_tile_config_args args;
+	uint32_t gpu_id;
+	HSAKMT_STATUS result;
+
+	result = validate_nodeid(NodeId, &gpu_id);
+	if (result != HSAKMT_STATUS_SUCCESS)
+		return result;
+
+	args.gpu_id = gpu_id;
+	args.tile_config_ptr = (uint64_t)config->TileConfig;
+	args.macro_tile_config_ptr = (uint64_t)config->MacroTileConfig;
+	args.num_tile_configs = config->NumTileConfigs;
+	args.num_macro_tile_configs = config->NumMacroTileConfigs;
+
+	if (kmtIoctl(kfd_fd, AMDKFD_IOC_GET_TILE_CONFIG, &args) != 0) {
+		return HSAKMT_STATUS_ERROR;
+	}
+
+	config->NumTileConfigs = args.num_tile_configs;
+	config->NumMacroTileConfigs = args.num_macro_tile_configs;
+
+	config->GbAddrConfig = args.gb_addr_config;
+
+	config->NumBanks = args.num_banks;
+	config->NumRanks = args.num_ranks;
+
+	return HSAKMT_STATUS_SUCCESS;
+}